Virtual AI4SE 2023: Generative Artificial Intelligence: The Future of Model-Based Systems Engineering

Model Based Systems Engineering (MBSE) has revolutionized the traditional engineering process by providing a digital and visual representation of systems and their interactions. This approach led to improved communication, reduced errors, and enhanced understanding of complex systems. Generative Artificial Intelligence (AI), particularly Large Language Models (LLMs), has the potential to further enhance MBSE by automating the generation of requirements, use cases, and activity diagrams in Systems Modeling Language (SysML). The objective of this paper is to apply AI for Systems Engineering (AI4SE) to create LLMs for our Government customers that streamline the systems engineering workflow automating the mundane tasks instead allowing systems engineers to focus collaboratively on development of complex system interactions. More time spent defining thorough system behaviors will yield comprehensive models ready for implementation by hardware and software engineering teams reducing the holes often found in today’s MBSE models for complex systems.

The integration of Generative AI and MBSE provides a unique opportunity to streamline the engineering process by leveraging the capabilities of LLMs. This in turn may allow systems modelers to do the following. 

• Automate the generation of requirements: LLMs can analyze existing requirements and generate new ones based on patterns and trends within the data, reducing the time and effort required for manual requirement generation. This enables Systems and Requirements engineers to focus on the true intentions of the requirements for the systems and sub-systems being developed.  
• Create use cases: LLMs can generate use cases by understanding the context of the system and its intended functionality, providing a more comprehensive understanding of the system's purpose and potential applications.
• Develop activity diagrams: LLMs can automatically generate activity diagrams that represent the flow of activities within a system, ensuring a consistent and accurate representation of the system's behavior.

Reducing Errors and Enhancing Robustness

The integration of Generative AI in MBSE can lead to a reduction in errors and the creation of more robust systems engineering models. LLMs can analyze existing data to identify inconsistencies and redundancies, resulting in improved model quality and a more accurate representation of the system. 

Furthermore, the use of LLMs allow system modelers to focus on the interactions within the MBSE models, leading to a deeper understanding of the system's behavior and potential issues. This focused approach can result in the identification and resolution of problems earlier in the engineering process, ultimately saving time and resources.

The integration of Generative AI, particularly LLMs, with Model Based Systems Engineering has the potential to revolutionize the engineering process by automating the generation of requirements, use cases, and activity diagrams in SysML. Future research should focus on the development and evaluation of LLMs specifically designed for MBSE integration, as well as the exploration of other AI techniques that can further enhance the MBSE process.

The presentation will focus on using current modern off the shelf Generative AI tools such as Google Bard, OpenAI ChatGPT, and others demonstrating the capabilities today of LLMs which have not been specifically trained and fine-tuned for the MBSE use case but can contribute greatly to augment a MBSE engineer’s workflow. 

References

1. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... & Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30.
2. Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., ... & Lowe, R. (2022). Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35, 27730-27744.
3. OpenAI. (2020). GPT-3: Generative Pre-trained Transformer 3
4. Google, Bard, http://bard.google.com
5. OpenAI, ChatGPT, https://chat.openai.com/
6. "Model-Based Systems Engineering (MBSE) - SEBoK". https://sebokwiki.org/wiki/Model-Based_Systems_Engineering. Retrieved 2023-06-01.